Professional Documents
Culture Documents
Fed M2 Seasonality
Fed M2 Seasonality
Fed M2 Seasonality
The emphasis placed on money stock changes end holiday buying. Housing starts in March
in the conduct and analysis of monetary reflect the beginning of favorable weather for
policy has recently increased interest in the building. Seasonal adjustments are based on
seasonal adjustment of the money stock. This an assumption that such influences are
new interest can be seen several ways. Money regular enough to be represented by calen-
market analysts, for example, are using an ap- dar months.
parent seasonal pattern in the seasonally ad-
justed money stock as an aid to prediction. The seasonal pattern
Also, annual revisions in the seasonal adjust-
ment factors the Federal Reserve uses in ad- Even casual examination of changes in
justing raw money figures seem to be getting the money stock shows a pronounced
bigger. The most recent revisions were the seasonal pattern. This is true regardless of the
largest ever. measure of the money stock. But it is most
These developments suggest there could pronounced for the narrowest measure—M1
be problems with the current procedure for (currency plus demand deposits)—on which
seasonally adjusting the money stock. As a this article focuses.
result, the Federal Reserve has appointed a
commission of outside economists and stat-
isticians to review the seasonal adjustment Unadjusted money data show
techniques used by the Board of Governors in a seasonal pattern
adjusting financial data. monthly percentage change in Ml *
This article examines the effect of 4—
12 Economic Perspectives
The money stock contracts for the first divided by the average monthly figure for the
two months of the year, and then expands year surrounding that month. The resulting
through the April tax date. It contracts in May, series of monthly figures, called the seasonal
then expands through June and July. August irregular ratios, shows whether the compo-
shows a mild contraction before an expansion nent tends to be high or low that month. The
that culminates in the year-end holiday high level of demand deposits in December,
season. for example, is reflected in seasonal irregular
The seasonal adjustment of money is ratios that from 1968 through 1977 were
complex. A simplified description of the ad- always above 1.02. Over that period,
justment is useful in explaining the effects— December demand deposits were always at
and problems—of the adjustment process. 1 least 102 percent of the surrounding 12-
The new money-stock figure published every month average.
month is adjusted by dividing previously Monthly seasonal irregular ratios form
determined seasonal adjustment factors into the base for the seasonal adjustment factors.
the raw (unadjusted) figures to obtain the Since the seasonal irregular ratios are ob-
seasonally adjusted money stock. Thus, a tained by dividing the unadjusted monthly
seasonal adjustment factor of 1.02 indicates figure by the average for the surrounding 12
the effect of that particular month is expected months, there is an implicit adjustment for
to raise the unadjusted data 2 percent above growth in the money stock. Also, the seasonal
the annual average. irregular ratio is based on an assumption that
The predetermined seasonal adjustment the effects of different months can be ex-
factors used to adjust data when they are first pressed as percentage changes in the unad-
published are the most current set of justed figure. If seasonal effects were assumed
factors—usually those used to adjust the constant—regardless of the size of the money
previous year's money stock. The seasonal ad- stock—then the 12-month surrounding
justment factor used to adjust the first- average would be subtracted from the unad-
published figure for November 1978, say, will justed figure.
be the same factor used in the 1978 revision of
the November 1977 figure. The factor used to The effect of changing seasonals
adjust a particular month's figure is revised
annually for several years after the first Seasonal adjustment factors can be ob-
published figure appears. In this article, the tained from seasonal irregular ratios in several
first published seasonally adjusted figure for a ways. One consideration is whether the
particular month is taken to be the first figure seasonal adjustment factors are assumed con-
for that month appearing in the Federal stant or whether they are assumed to change
Reserve Bulletin. over time. If they are assumed constant, equal
In the adjustment of Ml, separate weight is given to the same month every year.
seasonal adjustments are made to currency If the factors are believed to change over
and demand deposits and the two com- time, more weight is given to seasonal
ponents are then combined to obtain the irregular ratios in nearby years.
seasonally adjusted money stock. Unadjusted Current seasonal adjustment of the
monthly figures for each component are money stock assumes that seasonal factors are
changing. The rationale for a moving seasonal
is that the economic forces for which the
'For a more comprehensive description of the seasonals serve as proxy may be changing.
seasonal adjustment process, see Thomas A. Lawler,
"Seasonal Adjustment of the Money Stock: Problems and The importance of some holidays—or vaca-
Policy Implications," Economic Review, vol. 63: no. 6, p. tion habits—may change slowly over time,
23, (November-December, 1977), Federal Reserve Bank changing seasonal adjustment factors.
of Richmond. This article and its companion article dis-
cuss some of the issues raised here and list references to The extent to which the seasonal for any
other works on the same topic. calendar month is allowed to vary over
sharply changing factors that, while known to . Ddld covers 1968-1977 inclusive.
14 Economic Perspectives
revision, every one of the 17 large increases percentage point.
had been reduced and 15 of the 16 declines Clearly, revisions in the seasonal adjust-
showed smaller declines. Only seven in- ment sharply reduce the dispersion of the
stances of money declines remain in the most monthly percentage changes in the money
recent revision. supply—and the revisions are significant. The
Two factors account for the difference average absolute change produced by
between the first-published data and later seasonal revisions was 0.220 percentage
revisions. One is changes in the underlying point—nearly half the mean percentage
raw data. The other is changes in seasonal ad- change before seasonal revision.
justment factors. Separation of the effects of
these two factors make it clear that almost all A problem for analysis
the smoothing comes from seasonal revisions.
From 1968 through 1977, the first- The pronounced smoothing that comes
published data show a mean monthly change from seasonal revisions raises several
in money of 0.461 percent. A measure of the questions about the seasonally adjusted
dispersion in the data is the average absolute money series. Aside from whether the first-
deviation from this mean value—the average published data or the later revisions reflect
difference (up or down) between each obser- seasonal adjustments more accurately,
vation and the average observation. The smoothing poses problems of consistency for
average absolute deviation of first-published analysts of monetary policy.
data is 0.376 percentage point. When the first- Most research that relates changes in the
published data are adjusted to reflect money stock to changes in economic activity
revisions in the raw data, the mean monthly uses the smoothed seasonally adjusted
change rises to 0.494 percent and the average money series. Though seasonal revisions do
absolute deviation falls slightly to 0.364 not usually change the general pattern of
percentage point. When these data are acceleration or deceleration in growth of the
further adjusted to reflect the 1978 revisions money stock, the revised series is con-
in seasonal adjustment, the mean monthly siderably smoother than the first-published
change becomes 0.490 percent and the data. By relating this smoothed series to
average absolute deviation falls to 0.267 variations in economic activity, analysts
. . . are not greatly affected by revisions in . but are considerably smoothed by revisions
the underlying raw data .. . in the seasonal adjustment factors
number of observations number of observations
16 — 16 —
14 14
12 — 12
10 — 10
8 8
.54131 1 1 1 , 6 1 91,1 ,1 „ 1,
6 6
4 4—
2 — 2 L
01 o_
—10 .9 .1
monthly percentage change in M1. monthly percentage change in M1"
• Data covers 1968-1977 inclusive. •Data covers 1968-1977 inclusive.
16 Economic Perspectives
changes in the money stock. rates over the year but intensify the seasonal
Suppose that for two or three successive variation in money.
years monetary authorities ran an easy or tight While it is usually better to leave the
policy the same month, producing large or effects of policy in the seasonally adjusted
small rates of growth in money. Clearly, the numbers, policy effects producing a regular
policy effects should be kept in the seasonally seasonal in money seem different. If year in
adjusted money stock series. But unless some and year out, policy produces a seasonal
adjustment is made to take account of these movement in the money stock, it would seem
policy effects, the seasonal revision process is better to remove the policy effect from the
such that policy effects would be treated as a seasonally adjusted numbers. The difficulty is
shift in seasonal factors. They would be in distinguishing this recurring seasonal
removed from the revised seasonally adjusted pattern from the ephemeral effects of con-
data. Then, in later years when money tracyclical policy that happen to show some
figures revert to normal, the first-published seasonal pattern over a period of years.
data will be made more volatile by the interim
change in the seasonal. The importance of Summary
policy makes the judgmental contribution to The importance of money and the in-
seasonal adjustment of the money stock an creasing volatility of the money stock have
important factor. Effects of policy could, in focused attention on the seasonal adjustment
fact, explain the appearance of seasonals in of the money stock data. There is clearly a
seasonally adjusted data. pronounced seasonal in the money stock. It is
Policy considerations complicate not clear, however, exactly what the pattern is
seasonal adjustment in another way. To a or how it changes over time.
great extent, the seasonal variation in the One important characteristic of the
money supply is itself a product of monetary current method of seasonal adjustment is the
policy. The seasonal movement is actually a pronounced smoothing in the rate of growth
change in the demand for credit. Before the in the money stock produced by successive
Federal Reserve was established, the seasonal revisions of the seasonal adjustment factors.
effect was felt primarily on interest rates and This smoothing creates a danger of incon-
not nearly as much on money. The year-end sistency in the application of relationships es-
increase in credit demands, for example, timated with revised data to the analysis of
resulted largely in higher interest rates. The current monetary policy.
increase in money was only slight. More important, it raises the question of
Since the Federal Reserve was establish- whether the smoothing process produces
ed, it has acted to dampen the seasonal move- better seasonally adjusted data. The answer
ment in interest rates. This is done by in- involves serious problems of methodology as
creasing reserves and money when seasonal well as economics. Finally, the role of
demands for credit rise and reducing them monetary policy in determining the money
when credit demands fall. In this way, the stock makes judgmental modifications of
monetary authorities smooth out interest seasonal adjustment factors important.